Function nvfp4_scale_bf16 

pub fn nvfp4_scale_bf16(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    data: DevicePtr,
    scale: f32,
    total: u32,
    stream: u64,
) -> Result<()>

2026-09-25: data[i] *= scale in place over total BF16 values; the caller passes the down projection’s weight_scale_2.